Skip to content

Commit 8e2b658

Browse files
WanliZhongWanliZhong
authored
add person detection model from MediaPipe (#147)
1 parent 4126ce0 commit 8e2b658

File tree

10 files changed

+2779
-0
lines changed

10 files changed

+2779
-0
lines changed

‎README.md

+1
Original file line numberDiff line numberDiff line change
@@ -42,6 +42,7 @@ Guidelines:
4242
| [YoutuReID](./models/person_reid_youtureid) | Person Re-Identification | 128x256 | 30.39 | 625.56 | 90.07 | 44.61 | 5.58 | --- |
4343
| [MP-PalmDet](./models/palm_detection_mediapipe) | Palm Detection | 192x192 | 6.29 | 86.83 | 83.20 | 33.81 | 5.17 | --- |
4444
| [MP-HandPose](./models/handpose_estimation_mediapipe) | Hand Pose Estimation | 224x224 | 4.68 | 43.57 | 40.10 | 19.47 | 6.27 | --- |
45+
| [MP-PersonDet](./models/person_detection_mediapipe) | Person Detection | 224x224 | 13.88 | 98.52 | 56.69 | --- | 16.45 | --- |
4546

4647
\*: Models are quantized in per-channel mode, which run slower than per-tensor quantized models on NPU.
4748

‎benchmark/README.md

+8
Original file line numberDiff line numberDiff line change
@@ -95,6 +95,7 @@ mean median min input size model
9595
29.46 42.21 25.82 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
9696
6.14 6.02 5.91 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
9797
8.51 9.89 5.91 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb_int8.onnx']
98+
13.88 14.82 12.39 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
9899
30.87 30.69 29.85 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
99100
30.77 30.02 27.97 [128, 256] YoutuReID with ['person_reid_youtu_2021nov_int8.onnx']
100101
1.35 1.37 1.30 [100, 100] WeChatQRCode with ['detect_2021nov.prototxt', 'detect_2021nov.caffemodel', 'sr_2021nov.prototxt', 'sr_2021nov.caffemodel']
@@ -147,6 +148,7 @@ mean median min input size model
147148
762.56 738.04 654.25 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
148149
91.48 91.28 91.15 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
149150
115.58 135.17 91.15 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb_int8.onnx']
151+
98.52 98.95 97.58 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
150152
676.15 655.20 636.06 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
151153
548.93 582.29 443.32 [128, 256] YoutuReID with ['person_reid_youtu_2021nov_int8.onnx']
152154
8.18 8.15 8.13 [100, 100] WeChatQRCode with ['detect_2021nov.prototxt', 'detect_2021nov.caffemodel', 'sr_2021nov.prototxt', 'sr_2021nov.caffemodel']
@@ -200,6 +202,7 @@ mean median min input size model
200202
466.19 457.89 442.88 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
201203
69.60 69.69 69.13 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
202204
81.65 82.20 69.13 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb_int8.onnx']
205+
98.38 98.20 97.69 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
203206
411.49 417.53 402.57 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
204207
372.94 370.17 335.95 [128, 256] YoutuReID with ['person_reid_youtu_2021nov_int8.onnx']
205208
5.62 5.64 5.55 [100, 100] WeChatQRCode with ['detect_2021nov.prototxt', 'detect_2021nov.caffemodel', 'sr_2021nov.prototxt', 'sr_2021nov.caffemodel']
@@ -236,6 +239,7 @@ mean median min input size model
236239
1238.91 1244.87 1227.30 [640, 640] YoloX with ['object_detection_yolox_2022nov.onnx']
237240
76.54 76.09 74.51 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
238241
67.34 67.83 62.38 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
242+
56.69 55.54 48.96 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
239243
126.65 126.63 124.96 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
240244
303.12 302.80 299.30 [640, 480] DB with ['text_detection_DB_IC15_resnet18_2021sep.onnx']
241245
302.58 299.78 297.83 [640, 480] DB with ['text_detection_DB_TD500_resnet18_2021sep.onnx']
@@ -265,6 +269,7 @@ mean median min input size model
265269
1223.32 1248.88 1213.25 [640, 640] YoloX with ['object_detection_yolox_2022nov.onnx']
266270
52.91 52.96 50.17 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
267271
212.86 213.21 210.03 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
272+
221.12 255.53 217.16 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
268273
96.68 94.21 89.24 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
269274
343.38 344.17 337.62 [640, 480] DB with ['text_detection_DB_IC15_resnet18_2021sep.onnx']
270275
344.29 345.07 337.62 [640, 480] DB with ['text_detection_DB_TD500_resnet18_2021sep.onnx']
@@ -310,6 +315,7 @@ mean median min input size model
310315
428.66 524.98 391.33 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
311316
66.91 67.09 64.90 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
312317
79.42 81.44 64.90 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb_int8.onnx']
318+
84.42 85.99 83.30 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
313319
439.53 431.92 406.03 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
314320
358.63 379.93 296.32 [128, 256] YoutuReID with ['person_reid_youtu_2021nov_int8.onnx']
315321
5.29 5.30 5.21 [100, 100] WeChatQRCode with ['detect_2021nov.prototxt', 'detect_2021nov.caffemodel', 'sr_2021nov.prototxt', 'sr_2021nov.caffemodel']
@@ -387,6 +393,7 @@ mean median min input size model
387393
701.08 708.52 685.49 [1280, 720] DaSiamRPN with ['object_tracking_dasiamrpn_kernel_cls1_2021nov.onnx', 'object_tracking_dasiamrpn_kernel_r1_2021nov.onnx', 'object_tracking_dasiamrpn_model_2021nov.onnx']
388394
105.23 105.14 105.00 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
389395
123.41 125.65 105.00 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb_int8.onnx']
396+
134.10 134.43 133.62 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
390397
631.70 631.81 630.61 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
391398
595.32 599.48 565.32 [128, 256] YoutuReID with ['person_reid_youtu_2021nov_int8.onnx']
392399
1452.55 1453.75 1450.98 [640, 480] DB with ['text_detection_DB_IC15_resnet18_2021sep.onnx']
@@ -422,6 +429,7 @@ mean median min input size model
422429
20.62 22.09 19.16 [416, 416] NanoDet with ['object_detection_nanodet_2022nov.onnx']
423430
28.59 28.60 27.91 [640, 640] YoloX with ['object_detection_yolox_2022nov.onnx']
424431
5.17 5.26 5.09 [192, 192] MPPalmDet with ['palm_detection_mediapipe_2023feb.onnx']
432+
16.45 16.44 16.31 [224, 224] MPPersonDet with ['person_detection_mediapipe_2023mar.onnx']
425433
5.58 5.57 5.54 [128, 256] YoutuReID with ['person_reid_youtu_2021nov.onnx']
426434
17.15 17.18 16.83 [640, 480] DB with ['text_detection_DB_IC15_resnet18_2021sep.onnx']
427435
17.95 18.61 16.83 [640, 480] DB with ['text_detection_DB_TD500_resnet18_2021sep.onnx']

‎benchmark/config/person_detection_mediapipe.yaml

+19
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,19 @@
1+
Benchmark:
2+
name: "Person Detection Benchmark"
3+
type: "Detection"
4+
data:
5+
path: "data/person_detection"
6+
files: ["person1.jpg", "person2.jpg", "person3.jpg"]
7+
sizes: # [[w1, h1], ...], Omit to run at original scale
8+
- [224, 224]
9+
metric:
10+
warmup: 30
11+
repeat: 10
12+
backend: "default"
13+
target: "cpu"
14+
15+
Model:
16+
name: "MPPersonDet"
17+
scoreThreshold: 0.5
18+
nmsThreshold: 0.3
19+
topK: 1

‎benchmark/download_data.py

+4
Original file line numberDiff line numberDiff line change
@@ -213,6 +213,10 @@ def get_confirm_token(response): # in case of large files
213213
url='https://drive.google.com/u/0/uc?id=1LUUrQIWYYtiGoNAL_twZvdw5NkC39Swe&export=download',
214214
sha='4161a5cd3b0be1f51484abacf19dc9a2231e9894',
215215
filename='object_detection.zip'),
216+
person_detection=Downloader(name='person_detection',
217+
url='https://drive.google.com/u/0/uc?id=1RbLyetgqFUTt0IHaVmu6c_b7KeXJgKbc&export=download',
218+
sha='fbae2fb0a47fe65e316bbd0ec57ba21461967550',
219+
filename='person_detection.zip'),
216220
)
217221

218222
if __name__ == '__main__':

‎models/__init__.py

+2
Original file line numberDiff line numberDiff line change
@@ -8,6 +8,7 @@
88
from .face_recognition_sface.sface import SFace
99
from .image_classification_ppresnet.ppresnet import PPResNet
1010
from .human_segmentation_pphumanseg.pphumanseg import PPHumanSeg
11+
from .person_detection_mediapipe.mp_persondet import MPPersonDet
1112
from .qrcode_wechatqrcode.wechatqrcode import WeChatQRCode
1213
from .object_tracking_dasiamrpn.dasiamrpn import DaSiamRPN
1314
from .person_reid_youtureid.youtureid import YoutuReID
@@ -80,6 +81,7 @@ def register(self, item):
8081
MODELS.register(SFace)
8182
MODELS.register(PPResNet)
8283
MODELS.register(PPHumanSeg)
84+
MODELS.register(MPPersonDet)
8385
MODELS.register(WeChatQRCode)
8486
MODELS.register(DaSiamRPN)
8587
MODELS.register(YoutuReID)

0 commit comments

Comments
 (0)